CNR - Institute of Neuroscience CNR
Institute of Neuroscience
 

Project

Biodata analysis and decryption

Introduction

Development of computerized approaches to deal with biological issues has provided access to an incredible amount of scientific data. The informatic revolution is now spreading from molecules to subcellular, cellular and tissutal data through analysis of digital images of biological structures.

Although evident, self-explaining pieces of information can be readily extracted from images, a huge amount of information remains unexploited, mainly because of lack of expertise in data mining and biostructure analysis.

The goal of this project is to develop original computerized approaches to quantify elusive, but biologically significant, parameters. In addition, automatic routines will be used to set up high-throughput analyses of biological in support to research groups dealing with image analysis issues. In this respect a number of collaborations are currently under way.

Parametrization of blood vessel distribution

Anti-angiogenic treatments are promising approaches in the anti-cancer battle. However, experimental approaches currently used to quantify the activity of antiangiogenic treatments do not generally address the importance of spatial distribution of microvessels in target tissues. To quantify this parameter we devised a general computerized method to assess the space-filling property of a biological network. Our approach is based on pixel dilation of digital images of network elements (blood vessels, neurites, etc.) and is practically achieved computing the number of dilation cycles (Halo index) needed to permeate a pre-defined amount of each image. So far, our approach was validated on human tumour xenografts in nonobese diabetic/severe combined immunodeficient mice treated, or not, with the antiangiogenic drug sorafenib. For each experimental model, area normalization allowed the unbiased comparison of several hundreds of images showing different amounts of vascular tissue. In different tumour types, comparison of Halo values showed statistically significant differences between control and sorafenib-treated samples. Conversely, this effect was not observed in samples from xenografts known to resist the antiangiogenic treatment. By separating the number of network elements from the quantification of their distribution, our approach can thus contribute to a better description of the state of a network, highlighting subtle changes in element tortuosity, clustering or branching.

 

This methodological approach can prove helpful also in neurobiology as genetic defects can cause abnormalities in neurite networks and correlate with neuropathological disorders. This is the case, for example, with stau1 knock-out mice which show locomotory deficits. Similarly, changes in neurite arborization can be observed in Rett's syndrome. The availability of a method to quantify the space-filling property of a neurite layout can thus help in describing the extent of alterations observed in pathological samples. In addition, the same approach could help in quantifying the activity of new drugs, or treatments, in reverting the phenotypes.

Identification of synaptic vescicles in electron micrographs of presynaptic terminals

 

Synaptic vescicles are a characteristic feature of presynaptic terminals. Their numbers and position provide clues about the physiopathological state of the terminals for example in genetically modified animals. Although few images are sufficient to highlight evident defects, it is necessary to build up a large dataset of results to identify marginal differences. In such a case, spatial localization of hundreths of vescicles by hand appears to be a biased, time-consuming and error-prone task.

To solve this issue we are trying to develop a computerized method to automatically localize the center of synaptic vescicles using Fourier transforms and rotational invariant filters. Preliminary results highlight difficulties in the identification of poorly defined vescicles, although the routine appear adequate in removing signals due to dense-core vescicles, or artefacts occouring by chance.

Quantification of protein accumulation at exit sites of the endoplasmic reticulum (ER)

The export of proteins from the ER is a sorting process that takes place at defined structures termed exit sites (ERES).

Exported proteins appear to be sorted mainly because of their physicochemcal properties. Thus, to identify which features promote the export of known proteins it is necessary to quantify their distribution among ER, exit sites and downstream cellular compartments.

This analysis can be conducted by confocal imaging given that a) ERES can be identified, using specific markers and b) cells can be induced to synthetize engineered fluorescent reporter proteins that accumulate, or not, at ERES upon a low temperature block (10°C).

To automate the quantification of the fluorescent signals, we are developping an ImageJ routine that maps ERES and ER on the basis of their fluorescent staining. Then, for each ES, the routine looks for reporter proteins and quantify the extent of colocalized signals taking in account small spatial drifts. Results are then compared with the integrated density of reporter signals obtained from non-ERES regions of nearby ER. Finally the distribution in the ERES compartment is calculated according to the formula: 100%*(ERES / ERES+(nearby ER)). Actually we are validating the usefulness of this routine in CV1 cells using the ts045 mutant of the VSV reporter glycoprotein, at both permissive and non-permissive temperatures.

General research support

As a last point this group develops and manages research-oriented intranet services open to institutional Laboratories.

Publications

  • Righi M, Giacomini A, Lavazza C, Sia D, Carlo-Stella C, Gianni AM (2009) A computational approach to compare microvessel distributions in tumors following antiangiogenic treatments. Lab. Invest. 89:1063-70.

Collaborations

  • C. Carlo-Stella, Fondazione IRCSS Istituto Nazionale Tumori, Milan, Italy.

 

PI photo

Marco Righi

Contact information

email  E-mail

email  +39 02 5031 6969

Participating staff
[--]